Political Party Identification and Romantic Relationship Quality

Abstract

Although there are studies on voting similarity in mate selection, very little research has examined party identification and relationship quality. We assessed associations between reports of personal and partner voting behavior (Democrat, Republican, or Independent) and both relationship adjustment and commitment to one’s partner. We used a national (U.S.) sample of individuals (N = 510) in their 20s and 30s who were mostly in different-sex, serious relationships (Mdn duration = 5.25 years; 50% married). Controlling for age, college graduation, income, religiosity, minority status, and duration of relationship, voting Republican was associated with higher levels of commitment compared to voting Democrat. Similarly, being partnered with someone voting Republican was associated with higher relationship adjustment and commitment compared to being partnered with someone voting Democrat. Notably, differences in commitment were found only among those who were not already married or engaged. Findings were largely consistent with numerous online reports of survey data that have shown modestly greater relationship quality among those who report being politically conservative. While research has shown that people are less attracted to those who vote differently and that voting differently can be associated with lower relationship quality, we did not generally find perceived voting similarity to be associated with relationship adjustment or commitment. There was one exception: In exploratory analyses, Republicans paired with Democrats tended to report substantially lower relationship adjustment than if paired with Republicans. Implications for political differences among intimate partners and clinical work around these themes are discussed.

Political conversations dominate news coverage, social media, and the dinner table. Studies have examined the ways in which political preferences are related to various aspects of life, including romantic relationships, with questions surrounding political party identification being at the heart of much of the existing interdisciplinary research. The present study sought to examine associations between political party identification and relationship quality (commitment and overall relationship adjustment).

A number of studies have examined political similarity among partners (Klofstad et al., 2013; Iyengar & Westwood, 2014), providing evidence of selection for similarity (homogamy) on political identification. Typically, like ends up with like. For example, a recent national survey in the U.S. found that 77% of married and cohabiting partners reported that their partner belonged to the same political party (Pew, 2018). There has not been much evidence to suggest that political similarity weighs heavily in partner selection, though there is evidence that people shift toward liking a potential partner less when they discover dissimilarity in political orientation (Mallinas, Crawford, & Cole, 2018). In general, it seems that people gravitate towards partners who are similar to them on a variety of demographic dimensions (Coffé & Need, 2010; Klofstad et al., 2013), and this sorting, in turn, makes political similarity more likely. Other studies of similarity have focused less on matching than on how romantic relationships influence political leanings over time (e.g., Struber, 2010), with some evidence showing bi-directional influence between partners (Kan & Heath, 2006).

Throughout the rest of this paper, where we refer to the specific political parties of Democrat, Republican, and Independent, we describe the identification of U.S. voters. Further, in the U.S., Democrats tend to be more liberal and Republicans tend to be more conservative. Independents are heterogeneous, as they tend to span from liberal to conservative.

Attitudes towards inter-party marriage may be changing in today’s partisan climate. In an article for Voice of America, Kevin Enochs (2017) summarized data from 1958 and 2016 on how parents would feel if their child married someone from the opposite party. Gallup polling data from 1958 revealed that 33% of Democrats and 25% of Republicans wanted their child to marry someone from the same party. Enochs reported that political scientist Lynn Vavreck asked a similar question in 2016, finding that 60% of Democrats and 63% of Republicans wanted their child to marry someone from the same political party. Times have changed.

One study examined voting dissimilarity and relationship quality after the highly contentious election of President Trump in 2016. Afifi, Zamanzadeh, Harrison, and Torrez (2020) assessed married and cohabiting individuals before and after inauguration. They found that voting differently from one’s partner was associated with higher levels of stress, greater relational conflict, and a weaker communal orientation (e.g., being a team) among partners. Dissimilarity in voting may be an important and increasing stress on relationships.

While studies such as those noted above examine similarity as a primary focus, only a few, unpublished studies have addressed associations between politics and relationship quality. Wilcox and Wolfinger (2015) used the General Social Survey (2010–2014) to study differences in marital happiness based on political affiliation. They found that Republicans tended to report greater marital happiness than Democrats. About half of this difference was explained by variables other than political affiliation. Specifically, they found that Republicans are more likely to be white and more likely to be religiously involved, two factors independently associated with marital quality and stability (Broman, 2005; Bramlett & Mosher, 2002; Mahoney, 2010). However, Wilcox and Wolfinger reported that some difference between Republicans and Democrats remained even when controlling for such variables.

A large national survey in 2018 was used to examine the experiences of Americans in their relationships, marriages, and families, and included questions about how respondents typically voted and whether or not their political leanings were conservative, moderate, or liberal (Karpowitz & Pope, 2018). One of the questions asked was relevant to the topic of this paper. Among those in a relationship, respondents were asked if, at some point in the past two years, they had felt like their relationship was in trouble. Karpowitz and Pope (2018) reported that 34% of liberals, 35% of moderates, and 25% of conservatives said “yes.” This finding is consistent with those just noted by Wolfinger and Wilcox (2015), suggesting higher relationship quality among those with more conservative political leanings.

Wolfinger (2017) investigated the association between political party identification and both sexual frequency and fidelity. Although this study is not published in a journal, Wolfinger used a large U.S. data set and strong methods consistent with the field of sociology, finding that Republicans reported having more sex than Democrats, and that Democrats were more likely to report cheating on their spouse than Republicans. Additionally, this study indicated that individuals who identified as Independent were more likely to report infidelity than Republicans and similarly likely to report it as Democrats. Wolfinger further found that marital status explained the differences in sexual frequency, as Republicans are more likely to be married, and married individuals report higher sexual frequency. Controlling for a variety of other variables did not explain the observed differences between Republicans and Democrats in reports of infidelity. Even net of control variables, Republicans were 23% less likely to report cheating on their partner than Democrats.

Another investigation into the association between sexuality and voting behavior examined differences in sexual satisfaction by political identification. Using responses from more than 19,000 people from France, Germany, Sweden, Denmark, and the UK, a YouGov Survey (Twyman, 2016) found that conservatives reported being happier with their sex lives than liberals. Specifically, 71% of participants who identified as “very right-wing” reported being happy with their sex lives while 62% of participants who identified as “very left-wing” reported the same. Although the discrepancies (which did not exist in France) can be attributed to a number of different factors, the findings are consistent with the basic tendency for conservatives to report higher levels of relationship quality than liberals (e.g., Wilcox & Wolfinger, 2015).

It is clear that the existing literature on political identification and relationship quality is sparse. The present study tested whether political identification (Democrat, Republican, or Independent) was associated with relationship adjustment and commitment in romantic relationships of those who were either seriously involved or married. This study is important because, to our knowledge, there are no peer-reviewed, published studies on this question. Many factors associated with relationship quality have been examined over the years, but politics has been little studied in this regard. Given the high salience of politics in our society and the comparative dearth of published research on this subject, this study makes an important contribution to the field.

In this study, we examined individuals who were in serious, different-sex, romantic relationships (married or unmarried, with a median duration of five years). The data were obtained from the final wave of a national longitudinal sample (U.S.) from a project designed to study relationship development among individuals originally recruited in the age range of 18 to 34. While most of the sample is comprised of relationships of substantial duration, and 50% of the sample was married by the wave we analyze here, the sample does not reflect those who have been married for many years nor does it reflect those over 40 years of age.

Based on the sparse published literature that existed when we began this study, we did not have hypotheses about differences between the groups prior to conducting analyses, so we treated the analyses as addressing four research questions about possible group differences. We used past voting behavior as the indicator of political identification. Specifically, we tested whether how one typically voted (Democrat, Republican, or Independent) was associated with differences in commitment and relationship adjustment. We also examined whether there were differences based on peoples’ perceptions of how their partner typically voted. We also tested relationship status (being married or engaged) as a moderator in our analyses because marriage may represent a higher level of commitment and development in relationships. Further, because there has been a historical interest in the impact of voting similarity on relationships, we tested if perceived voting similarity was associated with the outcomes.

1. Is there a difference in relationship adjustment and commitment between the relationships of those voting Democrat, Republican, or Independent?

2. Is there a difference in relationship adjustment and commitment in relationships of individuals partnered with those voting Democrat, Republican, or Independent?

3. Does a commitment to marriage (being married or engaged) moderate the findings for voting or perceived partner voting?

4. Is perceived similarity in party identification between individuals and their partners associated with relationship adjustment or commitment to relationships?

The data used here are cross-sectional, and associations cannot be used to assert causality. Nevertheless, we believe that addressing these research questions in this sample advances the literature on voting behavior and relationship quality.

Method

Participants

Participants (N = 510) were individuals were drawn from a national (U.S.), longitudinal sample developed for a project on romantic relationship development. Details are available in (Rhoades, Stanley, & Markman, 2020). At the time of recruitment, all participants were in the age range of 18 to 34, and were unmarried but in a “serious, exclusive romantic relationship” of at least two months with a member of the opposite sex. The parent project was designed to study cohabitation prior to marriage. It began in 2006, before marriage was available in all states to same-sex couples, hence, only those in different -sex relationships were originally recruited. However, the project followed people through changes in relationships, and not all participants were in different-sex relationships at later time points (see below). Data were drawn from the 11^th wave (collected in 2011–2012) of the longitudinal sample because the questions about voting behavior were only asked at that time point.

The average age for participants this analytic sample was 29.67 (SD = 4.52, Range = 23 to 40 years of age). The modal income level was in the category of “$20,000 to $29,999,” ranging from under $4,999 to over $100,000. Fifty-eight percent (58%) had graduated from college with a bachelor’s degree. Regarding ethnicity, the sample was 8.1% Hispanic or Latino and 91.9% not Hispanic or Latino. Regarding race, the sample was 81.5% White, 10.8% Black or African American, 2.6% Asian, and 1.4% American Indian; 3.5% reported being of more than one race and 0.2% did not report race. In terms of income as well as race and ethnicity, the parent sample recruited at wave one was comparable to the English-speaking population of the United States (for unmarried, similarly-aged adults). For the analytic sample from wave 11, participants reported being in their present relationship a median of 63 months (or 5.25 years). Seventy-seven percent (76.7%) were living together and 49.6% were married. Five participants were in same-sex relationships.

Procedure

To recruit participants for the larger project, a calling center used a targeted listed telephone sampling strategy to call households within the mainland United States. After a brief introduction to the study, respondents were screened for participation. To qualify, respondents needed to be between 18 and 34 and be in an unmarried relationship with a member of the opposite sex that had lasted 2 months or longer. Those who qualified, agreed to participate, and provided complete mailing addresses (N = 2,213) were mailed forms within 2 weeks of their phone screening. Of those who were mailed forms, 1,447 individuals returned them (65.4% response rate); however, 152 of these survey respondents indicated on their forms that they did not meet requirements for participation, either because of age or relationship status, leaving a sample of 1,294. Across time points, participants were paid $40 for completing the 30- to 45-minute surveys. All procedures were approved by a university institutional review board.

For the purposes of this study, data from the 11^th wave of the larger project were analyzed because that is the only time point where questions about voting history were asked (it was also the final wave in the project). Of the original 1294 participants in the study at the 1^st wave, 922 returned surveys at wave 11. Of the 922, 716 were in a relationship characterized as dating, engaged, or married at that time. We restricted analyses to those who reported both on their own as well as their partner’s voting behavior so that all analyses presented here are on the identical sample of participants, in order to facilitate comparison of findings across analyses. That is, this listwise deletion approach assured that findings across respondent voting versus partner voting, and also voting similarity, would be based on the same sample. Of those in a relationship at wave eleven, 645 reported how they typically voted and 622 reported how their partner typically voted, with 595 reporting on both variables. Of the 595, 583 had data on both dependent variables as well as the voting variables. Of those with missing data on one or more of the control variables, no single variable had missing data on more than 10% of the sample, so we elected to analyze the data set with complete case data. There were two participants who were multivariate outliers who were excluded from the analytic sample, resulting in a complete case analytic sample of 510 individuals. We also conducted exploratory analyses (not presented) wherein we imputed missing data on covariates using multiple imputation, which would allow for a sample of 581 individuals. The findings for the analyses based on multiple imputation were entirely consistent with those from complete case data (N = 510), and the latter is what we present.

Measures

Political party identification.

Party identification was operationalized by responses to two questions about voting: “How do you usually vote?” and “How does your partner usually vote?” Respondents could respond with either Democrat, Republican, or Independent. Breakdowns by voting response are provided in the first part of the results section.

Commitment.

Commitment was measured using the 8-item Dedication Scale from the Revised Commitment Inventory (Owen, Rhoades, Stanley, & Markman, 2011; Stanley & Markman, 1992). Example items are “I want this relationship to stay strong no matter what rough times we encounter,” and “I like to think of my partner and me more in terms of ‘us’ and ‘we’ than ‘me’ and ‘him/her.’” For each item, a scale of 1 to 7 was used, with 1 = strongly disagree and 7 = strongly agree. Various studies support the reliability and validity of this measure and its close variants (e.g., Dew, Uecker, & Willoughby, 2018; Rhoades, Stanley, & Markman, 2010). In this sample, Cronbach’s alpha (α) = 0.91. The mean of these items was used in this study, and higher scores reflect more commitment to one’s relationship and partner (M = 5.88, SD = 1.18).

Relationship adjustment.

Relationship adjustment was measured using the 4-item version of the Dyadic Adjustment Scale (DAS-4; Sabourin, Valois, and Lussier, 2005). This scale assesses happiness, thoughts about dissolutions, confiding in one another, and a general item about how well the relationship is going. Using the scoring suggested by Sabourin et al., the mean for this sample was 16.87 (SD = 3.61) with Cronbach’s alpha (α) = .85.

Control variables.

We controlled for a number of standard demographic variables because they could impact interpretation of results. Control variables included a number of variables already described, including age, income, education level (dichotomized as having a bachelor’s degree or not), and relationship length. Income was rated in 13 ranges which we elected to treat as one continuous variable for purposes of controlling for income. Additionally, we controlled for overall religiosity, which was measured with one item: “All things considered, how religious would you say you are?” (M = 3.87, SD = 1.85). Finally, we controlled for race, which was dichotomized as white, not Hispanic (73.9%) compared to all others (26.1%).

Moderating variables.

We tested if findings differed by relationship status. As explained in the introduction, we reasoned that this variable could make a difference. Respondents could indicate if they were presently married, engaged, had plans to marry their partners, or were dating. We know from other analyses in the parent sample that engagement is a much strong predictor of eventually marrying one’s partner than merely planning to marry the partner. Therefore, we included both those already married (50%) or engaged (9%) in one group (59%) and those who were not engaged (plans or dating) in a comparison group (41%). We also examined for differences based on gender.

Results

We tested for multivariate outliers using Mahalanobis distance based on all of the variables used in the analyses to flag outliers within each voting group. This procedure identifies those who were in the extreme tail of a combination of all the variables used in the analyses (p < .001). One outlier was removed from among those typically voting Democrat and one outlier was removed from among those typically voting Independent.

Regarding their own voting, respondents reported voting as follows: 52.0% Democrat, 25.9% Republican, and 22.2% Independent. Respondents reported that their partner usually voted as follows: 51.2% Democrat, 29.6% Republican, and 19.2% Independent.

Regarding the control variables, respondents’ voting behavior was not significantly associated with differences in age or how long they had been in their relationships. However, the groups did differ in the percent graduating from college, income, religiosity, and minority status, in unsurprising ways. Post-hoc tests revealed that Independents were less likely to have graduated from college than Democrats (p < .01), with no significant difference between Republicans and Democrats. Both Republicans and Democrats reported greater income than Independents (p < .01 for both), with no significant difference between Republicans and Democrats. Republicans reported being more religious than either Democrats or Independents (p < .01 for both). Democrats were less likely to be white than either Republicans or Independents (p < .01, p < .05, respectively); Republicans and Independents were not significantly different in the likelihood of being white. Table 1 presents correlations between relationship adjustment, commitment, and the control variables (along with means and standard deviations for all variables).

Table 1.

Correlations

	Mean or %	SD	1	2	3	4	5	6	7
1. Relationship Adjustment	16.87	3.61	--
2. Commitment	5.87	1.18	.73**	--
3. Age	29.67	4.52	−.17**	−.10*	--
4. Bachelor’s degree	58.2%	--	.14**	.05	.01	--
5. Income	5.79	3.09	.07	.07	.28**	.36**	--
6. Religiosity	3.87	1.85	.09*	.11*	−.07	.02	−.08	--
7. White	74.0%	--	.08	.09*	−.08	.02	.00	−.13**	--
8. Relationship duration (mo)	67.08	44.00	.03	.15**	.19**	.10*	.04	−.02	−.01

Note: The mean for income corresponds to the category $20,000 to $29,999.

^*p < .05

^**p < .01

All primary analyses reported below were conducted with analysis of covariance (ANCOVA), comparing respondents on either relationship quality or commitment based on how they typically vote or perceived their partner to vote, with each analysis controlling for age, college graduation, income, religiosity, minority status, and duration of relationship. All post-hoc analyses were conducted with Bonferroni corrections to control for multiple tests, providing a conservative correction for post-hoc analyses (e.g., Field, 2013). Adding gender as a control variable did not change findings. Adding gender as a moderator, likewise, did not change findings except in a few instances where doing so yielded complex interactions that did not seem to us to tell a clear story. Thus, we do not address gender further in this paper.

Main Effects for Voting and Partner Voting

Our first research question was whether there would be a difference in the relationship adjustment and commitment levels of those voting Democrat, Republican, or Independent. Table 2 presents the adjusted cell means and results for the ANCOVA on relationship adjustment and commitment. There was no statistically significant difference in relationship adjustment based on how respondents voted, F(2, 501) = 2.11, p = .12. There was a significant difference in commitment based on how the respondent voted, F(2, 501) = 4.60, p = .01. Post-hoc tests indicated that commitment was higher for those usually voting Republican than for those usually voting Democrat, p = .008, d = 0.34.

Table 2.

Adjusted Cell Means and ANCOVA Results

		Adjusted Means (SD)
	Variable	Democrat	Republican	Independent	F-statistic
Vote	Relationship Adjustment	16.57 (3.77)	17.38 (3.43)	16.99 (3.30)	F(2,501) = 2.11
	Commitment	5.76 (1.21)	6.15 (1.03)	5.85 (1.16)	F(2,501) = 4.60*
Partner Vote	Relationship Adjustment	16.37 (3.97)	17.51 (3.03)	17.25 (3.14)	F(2,501) = 5.39**
	Commitment	5.70 (1.27)	6.11 (0.91)	6.01 (1.18)	F(2,501) = 6.40**

Notes. Controlling for age, bachelor’s degree or not, income, religiosity, white/non-white, and duration of relationship.

^*p < .05

^**p < .01.

Our second research question addressed if perceived partner voting would be associated with the respondents’ relationship adjustment and commitment to their partners. Table 2 presents the adjusted cell means and results for these analyses. There was a significant difference in relationship adjustment based on how respondents reported their partners usually voted, F(2, 501) = 5.39, p = .005. Post-hoc comparisons indicated that relationship adjustment was higher for those partnered with someone usually voting Republican than for those partnered with someone usually voting Democrat, p = .007, d = 0.31. Likewise, there was a significant difference in commitment based on how respondents reported their partners usually voted, F(2, 501) = 6.40, p = .002. Post-hoc comparisons indicated that respondents reported higher commitment if partnered with someone usually voting Republican than for those partnered with someone usually voting Democrat, p = .003, d = 0.35. As can be seen in Table 2, the means for either voting Independent, or perceiving one’s partner to vote Independent, tended to fall in between the means for Republican and Democrat voting.

Testing Moderation of Main Effects by Relationship Status

For research question number 3, relationship status (married or engaged vs neither) was tested as a moderator of the associations between voting behavior (of self or partner) and the outcomes. Because tests of interaction effects are inherently lower in power, we set alpha at .10 for these tests, and relaxed the significance level for the post tests to include trends.

There was no moderation by relationship status for the outcome of relationship adjustment, both for respondent voting, F(2,498) = .26, p = .770 and perceived partner voting, F(2,498) = .09, p = .912. However, there was moderation of both main effects noted earlier when commitment was the outcome. The association between respondent voting and commitment was moderated by relationship status, F(2,498) = 2.44, p = .089. Post-hoc analyses indicated that the commitment of Democrats to their partners was lower than that of Republicans among those who were not married or engaged, p = .041, d = 0.41, but that there was no difference between voting groups among those who were married or engaged. Likewise, the association between perceived partner voting and commitment to one’s partner was moderated by relationship status, F(2,498) = 3.32, p = .037. Among those who were not married or engaged, the commitment of respondents to their partners who they perceived to vote Democrat was lower than among those who perceived their partners to vote Independent, p = .002, d = 0.50. Similarly, among those who were not married or engaged, there was a trend for the commitment of respondents to their partners who voted Democrat to be lower than that among those whose partners voted Republican, p = .061, d = .36.

In summary, for all of these analyses testing moderation by relationship status, there were no statistically significant differences in commitment levels based on voting among those who were married or engaged, suggesting earlier effects for both personal voting and perception of partner voting on commitment are mostly driven by differences among those who are not married or engaged. In contrast, the pattern of respondents reporting lower relationship quality if in a relationship with a Democrat held regardless of relationship status.

Voting Similarity

Our third research question was whether differences between respondents’ own voting and their perception of their partners’ voting were associated with relationship adjustment or commitment. Table 3 displays the percentage of perceived voting similarity by voting group. Overall, seventy-five percent (74.7%) reported that they and their partner voted similarly, with Democrats reporting the highest similarity (81.9%), followed by Republicans (74.2%) and Independents (58.4%), χ2(2) = 23.13, p < .001. Analyses of covariance were used to test if voting similarity was associated with relationship adjustment or commitment. These analyses used the same control variables as noted earlier. We compared groups based on if the respondent reported that their partner voted the same as they did (74.7%) or differently (25.3%). In the overall analyses, there was no significant difference in reported relationship adjustment or commitment based on voting similarity, F(1, 502) = .38, p = .537, and F(1, 502) = 1.21, p = .272, respectively.

Table 3.

Perceived Voting Similarity

		Partner Voting
		Democrat	Republican	Independent	Row Total
Respondent Voting	Democrat	N = 217, 82%	N = 29, 11%	N = 19, 7%	N = 265
	Republican	N = 21, 16%	N = 98, 74%	N = 13, 10%	N = 132
	Independent	N = 23, 20%	N = 24, 21%	N = 66, 58%	N = 113

Note. Percentages represent proportion of total within rows.

These analyses of voting similarity by main effect are not sensitive to the type of finding noted earlier, where respondents reported lower relationship adjustment and commitment if they were partnered with someone voting Democrat. Thus, we conducted exploratory analyses that allowed us to look at all specific pairings using the same models as above, but specifically within each voting group apart from the others. For example, we explored effects for differences in relationship adjustment based on how one’s partner typically voted, within each level of how the respondent voted. It is important to note that these analyses are considerably less robust than the main effects for voting and partner voting presented earlier because of the smaller cell sizes involved in some comparisons (e.g., the number of those voting Republican who reported their partner usually voted Democrat was only 21 whereas 98 people voting Republican reported that their partner also voted Republican). As such, these analyses may not detect differences that actually exist or could yield differences that are not reliable. Hence, these analyses are considered exploratory.

As suggested by the analyses on voting similarity presented earlier (with all groups included at once), voting similarly or differently from one’s partner was not generally associated with either relationship adjustment or commitment in these exploratory analyses within voting category, but with one exception. Post-hoc tests suggested that among those who vote Republican, relationship adjustment was lower if they reported that their partner voted differently F(2, 123) = 5.55, p = .005. In particular, post-hoc comparisons showed that if a Republican’s partner voted Democrat, they reported lower relationship adjustment, p = .003, d = 0.79. There were no significant differences on commitment in these exploratory analyses, though the findings trended the same direction as for relationship adjustment, with both Republicans and Independents partnered with those voting Democrat tending to report less commitment to their partners. All together, these findings suggest that the significant differences for relationship adjustment noted earlier based on partner voting behavior are mostly driven by the fact that Republicans rated relationship adjustment lower if they were not in a relationship with another Republican, and particularly if in a relationship with a Democrat.

Discussion

Our sample was of those who are in serious, romantic relationships, including marriage or clear plans for marriage to one’s partner and in earlier stages of life (40 years of age and under). The findings are not necessarily generalizable to those who are older or in longer-term marriages. Further, our analyses are correlational and cannot establish a direction of causality or demonstrate, for example, what other variables may produce the associations we found. Nonetheless, the findings indicated significant differences between those voting Democrat or Republican in relationship quality.

Those typically voting Republican reported higher commitment to their partners than those voting Democrat, but there were not significant differences in relationship adjustment. At the same time, those partnered with Republicans reported higher relationship adjustment than those partnered with individuals voting Democrat. This finding was not moderated by relationship status (i.e., being married/engaged vs. dating). However, exploratory analyses suggested that this finding is most likely explained by Republicans, who reported lower relationship adjustment if they were in a relationship with a Democrat. That comparison resulted in the largest effect size of these analyses (d = 0.79), while the rest of the effects were small to moderate in size (ranging from 0.31 to 0.51).

The differences in commitment were moderated by relationship status, with the differences being statistically significant only for those who were not married or presently engaged. Among those who were married or engaged, which was 59% of our sample, there were no statistically significant differences, suggesting that Democrats were as committed as Republicans, and those partnered with Democrats were as committed as those partnered with Republicans. However, among those not married or engaged, Democrats scored lower in commitment, and those in relationships with Democrats also scored lower in commitment to their partners.

The context of this sample was a project examining individuals in the stage of life where many people are sorting out to whom, and when, to settle down in marriage. Overall, 59% were married or engaged, but were also at an earlier stage of life than many studies that have focused on political aspects of romantic relationships in the past. Our findings regarding commitment suggest that Republicans may be more generally committed to partners prior to marriage or engagement. For Democrats, levels of commitment may be conditional on having settled on a future in marriage with one’s partner, but it could also be that Democrats in serious relationships of longer duration (like the average relationship studied here) are simply less committed to their partners. These differences were obtained even though Democrats and Republicans were together with their partner similar lengths of time, and when controlling for various other demographic variables such as age, income, race and educational attainment. It may be that some Democrats in this sample who were not married or engaged are simply in less committed relationships even though those relationships are of substantial duration. Along the same lines, perhaps a greater number of the non-married, non-engaged Republicans will develop those plans in the future. Indeed, Republicans are more likely to be married (67%) than Democrats (45%) or Independents (52%) (Pew, 2010). It is clear in the findings presented here that, outside of marriage, Republicans and those partnered with Republicans reported higher levels of commitment to their partners than Democrats. Democrats may be less likely to marry because they are less committed to their partners in unmarried relationships, or they may be less committed to partners they are unlikely to marry. Given that these data are from the last wave in a longitudinal sample, we have no way of adjudicating among such possibilities.

Although there is a broad literature on how homogamy on various dimensions is associated with marital quality and stability (e.g., Hohmann-Marriott & Amato, 2008; Mahoney, 2010), we did not find similarity in voting to be associated with relationship adjustment or commitment, with the notable exception of Republicans, where we did find evidence that similarity in voting was positively associated with relationship outcomes. Republicans were most likely to report lower relationship adjustment if partnered with a Democrat. It may be that similarity in voting ideology and behavior is somewhat more important in the relationships of Republicans than Democrats or Independents. This finding should be considered tentative and in need of replication, being based on an exploratory analyses.

The sample used here is one where half the respondents were married. Perhaps voting homogamy would be more associated with relationship adjustment and commitment in a sample of married people who have been together longer. If so, causality could work in either direction in explaining such a finding; those who are more similar are more likely to stay married and similarity is more likely to emerge over time in committed unions. It is also possible that this sample is of a generation where similarity on such a dimension may not be as prized, though that runs counter to evidence that people have become increasingly polarized, where such differences would be anticipated to lead to further struggles. We should note as well that most people in the sample were in a relationship with someone sharing their voting identification, and we have no way to determine how many people had, prior to being in the present relationship, possibly dodged unhappiness by avoiding dating someone who tended to vote differently.

Taken as a whole, these findings are modestly consistent with other findings such as the YouGov report of higher sexual satisfaction for conservatives compared to liberals (Twyman, 2016), Karpowitz and Pope’s (2018) national survey and report showing conservatives being the least likely to report feeling like their relationship had been in trouble, and Wilcox and Wolfinger’s (2015) findings that Republicans tend to report greater marital happiness than Democrats. Further, that Republicans reported higher levels of commitment than Democrats, especially if there was no commitment in or to marriage, may be consistent with Wolfinger’s (2017) findings that Democrats were more likely to report infidelity than Republicans, net of numerous control variables, including relationship status. It is possible that we would have found a commitment difference as well, even among those married or having a commitment to marry, had we had a much larger sample, as the mean for Republicans was higher than the mean for Democrats in the group committed to marriage, though not statistically significantly so (d = .20). We should also note that the differences we found along with those found in other studies are consistent but they are not generally large. It is not like you know the destiny of a relationship if you know how each of the two partners vote.

One possible explanation for the discrepancy between commitment levels among Republicans and Democrats could be that conservatives have a self-reporting bias. Commitment is strongly emphasized in the value systems of conservatives. Republicans may also face greater pressure from others to demonstrate commitment. Thus, Republicans could be more committed, or they may simply report being more committed because of a strong internalization of associated values or because of social pressure. The finding that people with partners who typically vote Democrat are less dedicated to the relationship could extend from the same reasoning. Democrats reported lower levels of commitment than Republicans; so, if one’s partner is a Democrat, and if they are indeed demonstrating lower commitment, perhaps people will, in turn, be less committed to their partners. Though a possibility, this explanation of self-report bias is undermined by there being no inherent reasons to believe that Republicans have a self-reporting bias that would explain our findings. More importantly, these findings are entirely consistent with the direction of all of the other findings discussed. The literature is sparse but consistent.

Another possible explanation for the commitment findings derives from social psychology research conducted by Haidt (2012). Haidt, whose work centers on global conceptions and manifestations of morality, argues there are six themes or foundations to morality. The themes, which he identifies and labels as care, fairness, liberty, loyalty, authority, and sanctity, exist in varying forms and magnitudes in communities around the world. After John Kerry’s defeat in the 2004 presidential election, Haidt began to apply his Moral Foundations Theory, originally developed in 2002, to politics. In a study that polled over 130,000 individuals, Haidt attempted to determine how liberals and conservatives consider or utilize the six foundations of morality when making moral judgments. The results of his study show that while conservatives tend to use all six foundations relatively equally, liberals lean on care, liberty, and fairness and frequently disregard sanctity, loyalty, and authority.

It is not difficult to imagine how each of Haidt’s moral foundations could impact the workings and results of romantic relationships. For example, fairness and liberty within a romantic relationship could have major implications for overall relationship adjustment, but since both dimensions are animating to liberals and conservatives, they may not differentiate these groups regarding relationship behavior. On the other hand, Haidt finds that liberals tend to be less motivated by the “loyalty” foundation of morality, which may be relevant to our findings associated with commitment. It is conceivable that persons less concerned with loyalty may also be less dedicated to their partners.

Contemporary political science research may also illuminate the findings. In Prius or Pickup, Hetherington and Weiler (2018) used responses to questions about parenting attitudes to distinguish liberals and conservatives by worldview. According to them, liberals tend to have more “fluid” worldviews whereas conservatives have more “fixed” ones. Those with a fluid worldview tend to perceive the world as being full of opportunities that can and should be explored. As a result, people with fluid worldviews are more likely to be open to new experiences and non-traditional pathways. Conversely, those with more fixed worldviews tend to be skeptical about the safety of the world and its occupants, preferring that which is orderly, familiar, and comfortable. These differing worldviews may impact romantic relationships of Democrats and Republicans. For instance, there is a consistent body of research showing that those who are more committed are less likely to be monitoring for potential alternative partners (e.g., Johnson & Rusbult, 1989; Stanley, Markman, & Whitton, 2002). In relationships not already defined by marriage or engagement, Democrats may be less committed to their partners because they may be more likely to remain open to alternative options while Republicans may be more prone to closing off such possibilities at less committed stages of relationships.

In most all analyses, Independents tended to fall somewhere in between the Republicans and Democrats. That may not be surprising. In general, we do not believe that our findings are very informative about Independents because the subsample is the smallest and they are a heterogeneous group.

More research could be conducted on how political tendencies are associated with romantic relationship behavior. The field is at a very early stage. While findings thus far appear to have some consistency, there is too little research on too few dimensions to make any broad conclusions. More research with larger samples and a greater breadth of relationship types is needed to better understand voting behavior and relationship quality.

Clinical Implications

Nothing in our analyses or in reasoning about political identification and romantic relationships suggests that political ideology causes differential relationship outcomes. Any implications about the practice of therapy or relationship education lie less in trying to influence anyone politically and more in being alert to how such issues may impact people in their most important relationships. As often noted, we live in an era of increasing political partisanship and division. We can think of two specific implications, one to do with matching and one to do with management.

Couples who are successful are able to navigate all sorts of differences in the beliefs, tendencies, and behavior of the two partners. While love may not conquer all, love and commitment can surely overcome much. Among those who have not yet settled into a serious, long-term, committed relationship (particularly, in marriage), there might be reason for some caution about matching with a romantic partner and potential mate on political identification. That may be especially true for Republicans. Matching on political tendencies is intuitive to many, as people report that they would not consider marrying (or dating) someone of an opposite political persuasion; in fact, people report lower attraction to those with different political viewpoints (Mallinas et al., 2018). And yet, with the advent and ubiquitous use of dating apps and websites, it may be easier than ever for people to become interested and involved with others who do not share similarity in backgrounds and experiences that make similarity in political identification more likely than in the past. Should people be more careful? Should people vet potential dating partners more about their politics before taking any attraction forward? We are not optimistic about the likelihood of this happening on a wide basis. On the other hand, those working with individuals in therapy or relationship education do have opportunities to raise with their clients the potential importance of deciding, beforehand, on what dimensions they believe similarity to be most important to them.

Conversely, among couples who are already deeply engaged in a life together, perhaps in marriage and or having children together, what are two partners to do who find themselves not merely having political differences but being divided by such differences? The study noted earlier by Afifi et al. (2020) found that dissimilarity in voting might place a substantial load of stress on a relationship, especially in the highly partisan times in which we live. In contrast, we did not find broad differences in relationship quality by voting similarity, though it is worth noting that the data we used were collected in the year before the 2012 election, not in 2016. Afifi et al. also noted that those who reported regularly engaging in relationship maintenance behaviors faired best whether or not they reported voting similarity to their partner.

Consistent with the findings of Afifi et al., we would think that political differences between partners will be amenable to the same strategies that can help ameliorate other differences undermining the basis for a relationship, including clarification of expectations and values, good communication, conflict management, and setting boundaries on where and when (and if) politics will be discussed (e.g., Christensen, Doss, & Jacobson, 2014; Markman, Stanley, & Blumberg, 2010). In an increasingly partisan society, it will be important for couples with political differences that are causing conflict to learn ways to tolerate each other and protect the good things they have together from such areas of strong disagreement. The findings mentioned earlier in the report by Enochs (2017) suggest that couples dealing with political differences may also have to cope with strains between the partners and their families, including alienation from one’s own family and/or hostility from one’s family toward the partner. Such couples are likely to be helped by strategies designed to strengthen their commitment to one another and their shared identity on other values and behaviors. In other words, couples navigating political difference may need help being on the same team in life even if they are on different teams in the voting booth.

Limitations

This study has a number of limitations. First, all of the data are self-reported with no objective measurement of the key predictors or outcomes. Second, while the sample began as a large, national sample of individuals in serious romantic relationships that well reflects the make-up of people living in the United States at recruitment (Rhoades, Stanley, & Markman, 2010), our analyses could only be conducted within the 11^th wave, which was nearly five years following sample recruitment—a time during which the sample undoubtedly became more select for some characteristics. Most notably, while the original recruitment resulted in a sample where the average relationship had existed for just under two years, the sample here was comprised of individuals in relationships that averaged just over five years duration. The findings do not necessarily represent findings that would have been obtained within a sample of new dating relationships or of exclusively, long-term married relationships. This sample is somewhere in between on those dimensions. Third, the sample had quite a few more Democrats than Republicans or Independents. From the standpoint of statistical power, we would prefer a sample with greater numbers of the latter groups. On the other hand, the make-up of this sample tracks quite well with the shift to those born since 1981 being increasingly likely to be registered as Democrats or Independents compared to Republicans (Pew, 2018). There are a variety of tendencies for differences in these data that, if they remained of similar size, would be statistically significant with a larger sample. Fourth, the data are cross-sectional, with no ability to examine how voting behavior predicts longer term outcomes or how various relationships unfold. The data cannot speak to causality in terms of either voting behavior leading to changes in relationships or vice versa. Fifth, our sample is not large enough to examine various types of moderating factors that may well exist. For example, we did not report findings from analyses including gender as a moderator because, in the limited instances where differences emerged, we had no confidence that we had a large enough sample to interpret the resulting patterns. A much larger sample would allow for more fine-grained tests of various differences explored (or not explored) in this study. Sixth, as noted in the methods, the parent project was designed to study cohabitation prior to marriage, and the original sample was recruited in 2006, before marriage being available in all states to same-sex couples. While five participants in the analytic sample reported being in a same sex relationship, everyone reported being in a different-sex relationship at the start of the project. Hence, the sample does not well represent those in same-sex relationships and may not generalize to those in them. Given the paucity of published research on the subject of this paper, these limitations are countered by a strong, reasonably sized sample wherein voting behavior could be examined regarding relationship adjustment and commitment.

Conclusions

With an increasingly polarized environment, people are curious about differences in life outcomes and pathways based on political leanings. Overall, the findings here were quite consistent with the few findings previously reported that bear on the question of relationship quality as it is associated with political identification. We found consistent evidence for some edge for those with conservative leanings, favoring higher relationship quality. However, these differences were clearest for those in relationships among those who are not married or engaged. And yet, regardless of that moderating factor, those in relationships with Democrats generally reported lower relationship adjustment than those in relationships with Republicans. On the often discussed and written about matter of romantic relationship difficulties associated with voting dissimilarity, our primary analysis did not suggest that it mattered all that much. That being said, we did find evidence that voting similarity, when it does matter, may matter most for Republicans, but such findings should be replicated in other studies and larger samples before being accepted as generally true. There is much more that could be learned about voting patterns and the politics of love and family. This study makes an important contribution to a nascent literature, but there is clearly room for more systematic research on how political identification is associated with relationship outcomes. We would vote for that.